OpenShift Compliance Drift and Audit Correlation
Medium severity TLP:CLEAR Research by Elastic Security Labs Published 20 Nov 2026
Hypothesis
Unauthorized changes to OpenShift resources such as pods or service accounts trigger compliance violations that the hunt traces to a specific actor by correlating scan failures with Kubernetes API audit logs.
Brief
Context Elastic Security
Labs recently published an article titled From posture to root cause: Investigate OpenShift compliance with Elastic Agent Builder (https://www.elastic.co/security-labs/blog/openshift-compliance-elastic-agent-builder). The article describes how to move beyond simple posture alerts to find the root cause of compliance violations. This hunt translates those concepts into a structured playbook to automate the investigative flow by linking disparate data sources.
How the Hunt Flows
The hunt begins with a scoping phase that queries the hb_compliance_finding surface. This step identifies failed compliance controls reported by the OpenShift Compliance Operator, such as failures in CIS or NIST benchmarks. By starting with failed controls, the hunt ensures that the subsequent high-volume audit searches are targeted only at clusters with known issues. Next, the hunt employs a logic gate to assess the severity of these failures. An analyst reviews the findings to see if they impact sensitive resource types in production environments. This check prevents the fan-out of more expensive queries for development clusters or low-severity findings that do not justify a deep-dive audit investigation. If the violations are significant, the hunt proceeds to the correlation phase. The correlation phase runs two investigative paths in parallel. The first path queries the hb_cloud_api_activity surface for Kubernetes API operations, specifically looking for create or patch actions that occurred within the lookback window. This query targets the specific resources identified in the scoping phase. The second path examines hb_process_activity on the underlying nodes to find rare executions of tools like oc or kubectl. Interactive use of these tools on a worker node is a strong indicator of manual tampering that might not follow standard deployment workflows. In the triage phase, an analyst correlates the timing of these activities. By matching the timestamp of an API modification or a rare process execution with the timing of the compliance scan failure, the hunt identifies the specific actor or service account responsible for the drift. This correlation is what makes this a hunt rather than a simple detection rule, as it links data across compliance, cloud, and host telemetry surfaces to build a complete narrative.
What the Hunt Cannot See
There are several blind spots to consider. If the OpenShift Compliance Operator is disabled or fails to report, the lead query will show a clean environment even if drift exists. Furthermore, if Kubernetes audit logs are not configured with sufficient verbosity, the logs might record the API operation without the specific patch payload, requiring manual cluster inspection to see the exact change. Finally, visibility into interactive CLI tool usage depends on having the Elastic Agent installed on all worker nodes.
Steps
-
Query, scoping
Lead: Identify Failed Compliance Controls
Find clusters and specific resources that have failed compliance checks according to the OpenShift Compliance Operator.
sql, reads hb_compliance_findingSELECT resource_uid, resource_name, check_name, status, severity, framework_refsFROM hb_compliance_findingWHERE LOWER(provider) = 'openshift' AND LOWER(status) = 'fail'What a hit looks like. Rows indicate failed CIS or NIST controls. Silence provides evidence that the operator reports no active violations across the estate.
-
Agent triage
Initial Assessment of Failures
Evaluate whether the identified compliance failures represent a security risk worth investigating via expensive audit log analysis.
-
Decision
Gate: Proceed to Audit?
Route the hunt to expensive audit queries only when the compliance lead shows active, relevant violations.
-
Query, detection candidate
Kubernetes API Resource Modifications
Find recent create or patch operations on sensitive resources that likely caused the compliance drift.
sql, reads hb_cloud_api_activitySELECT time, actor_user_name, api_operation, resource_type, resource_name, cloud_account_uid, src_endpoint_ipFROM hb_cloud_api_activityWHERE provider = 'kubernetes' AND (LOWER(api_operation) = 'create' OR LOWER(api_operation) = 'patch') AND instr(',' || '{{sensitive_resources}}' || ',', ',' || LOWER(resource_type) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days')What a hit looks like. API activity identifying the actor who modified the resource. Signal is validated when it precedes the compliance scan failure.
-
Query, baseline
Rare CLI Tool Execution on Nodes
Identify rare interactive usage of cluster tools like oc or kubectl directly on nodes, indicating manual tampering.
sql, reads hb_process_activitySELECT LOWER(process_path) AS path, COUNT(DISTINCT device_hostname) AS host_count, MIN(time) AS first_seenFROM hb_process_activityWHERE (LOWER(process_name) = 'oc' OR LOWER(process_name) = 'kubectl') AND ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0) AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY 1HAVING host_count < 3What a hit looks like. A rare process execution path for a cluster CLI tool. Silence suggests normal automated operations.
-
Agent triage
Triage: Correlate Findings
Link the timing of the compliance failure with the specific API modification or rare CLI execution to identify the root cause.
-
Decision
Route to Remediation
Decide whether to revert the unauthorized change based on the triage verdict.
-
Response action
Revert Resource Modification
Use the OpenShift Compliance Operator or direct API commands to revert the unauthorized resource state.
-
Analyst task
Analyst Review
Review the actor's intent and verify that the remediation was successful.
-
Analyst task
Close Out
Document findings and determine if a permanent detection rule is needed for the observed drift pattern.
Coverage
3 of 3 stages covered
| Stage and ATT&CK technique | Covered | How, or why not |
|---|---|---|
| Unauthorized Cluster Resource Modification T1610 T1059.004 |
Yes | audit-api-modifications, rare-cluster-ops |
| Compliance Violation Detection T1613 |
Yes | compliance-drift-lead |
| Root Cause Investigation and Remediation T1562 |
Yes | triage-root-cause |
Blind spots
- Needs OpenShift Compliance Operator running and reporting to hb_compliance_finding. A failed scan process results in zero reported failures in the lead query, leaving drift completely unobserved. It would answer: Is the cluster passing, or is the compliance operator failing to report?
- Needs hb_cloud_api_activity with full Kubernetes audit verbosity. Audit logs that record the action but not the patch payload require live cluster inspection to confirm the drift cause. It would answer: What specifically changed in the resource patch that triggered the violation?
- Needs Elastic Agent coverage on every OpenShift node. Process execution on worker nodes remains invisible without node-level telemetry, masking interactive tampering. It would answer: Was a CLI tool executed interactively on a node without an agent?
Parameters and data
Parameters
| Parameter | Type | Default | What it is |
|---|---|---|---|
lookback_days | number | 14 | Days of history to examine for audit and process telemetry. |
scope_hosts | list[host] | none | Optional list of hostnames to focus the node-level process baseline. |
sensitive_resources | list[string] | pods, serviceaccounts, rolebindings, complianceremediations, clusterroles | Kubernetes resource types monitored for unauthorized modifications. |
Data sources
| Source | Category | Telemetry |
|---|---|---|
| Endpoint telemetry (hb_ surfaces) | endpoint | endpoint |
Source
An open hunt.md file. It runs anywhere that reads the format.
---
analysis: A standard rule alerts on a failed check; this hunt correlates findings
from three surfaces (Compliance findings, API audit, and node-level Process) to
identify the actor and timing, which no single rule can automate.
blind_spots:
- id: compliance-operator-down
question: Is the cluster passing, or is the compliance operator failing to report?
requires: OpenShift Compliance Operator running and reporting to hb_compliance_finding
risk: A failed scan process results in zero reported failures in the lead query,
leaving drift completely unobserved.
stage: compliance-violation-detection
- id: insufficient-audit-detail
question: What specifically changed in the resource patch that triggered the violation?
requires: hb_cloud_api_activity with full Kubernetes audit verbosity
risk: Audit logs that record the action but not the patch payload require live cluster
inspection to confirm the drift cause.
stage: unauthorized-resource-modification
- id: node-agent-coverage
question: Was a CLI tool executed interactively on a node without an agent?
requires: Elastic Agent coverage on every OpenShift node
risk: Process execution on worker nodes remains invisible without node-level telemetry,
masking interactive tampering.
stage: unauthorized-resource-modification
coverage:
- stage: unauthorized-resource-modification
status: covered
steps:
- audit-api-modifications
- rare-cluster-ops
- stage: compliance-violation-detection
status: covered
steps:
- compliance-drift-lead
- stage: root-cause-investigation
status: covered
steps:
- triage-root-cause
guardrails:
claims: no_unsupported
evidence: citation_required
missing_data: not_benign
telemetry: untrusted
hunt:
applicability: campaign-specific
handoff: promote-to-detection
justification: Compliance drift in regulated environments is a primary indicator
of cluster persistence; tracing it to an actor ensures security accountability.
methodology: model-assisted
trigger: intel-report
hypothesis: Unauthorized changes to OpenShift resources such as pods or service accounts
trigger compliance violations that the hunt traces to a specific actor by correlating
scan failures with Kubernetes API audit logs.
labels:
- hunt
- attack.t1610
- attack.t1059.004
- attack.t1613
- attack.t1562
- discovery
- execution
- impact
name: OpenShift Compliance Drift and Audit Correlation
parameters:
lookback_days:
default: '14'
description: Days of history to examine for audit and process telemetry.
from:
kind: manual
observed: '2024-11-19'
ref: hunt-standard-lookback
type: number
scope_hosts:
default: []
description: Optional list of hostnames to focus the node-level process baseline.
from:
kind: manual
observed: '2024-11-19'
ref: analyst-defined-scope
type: list[host]
sensitive_resources:
default:
- pods
- serviceaccounts
- rolebindings
- complianceremediations
- clusterroles
description: Kubernetes resource types monitored for unauthorized modifications.
from:
kind: article
observed: '2024-11-19'
ref: https://www.elastic.co/security-labs/blog/openshift-compliance-elastic-agent-builder
type: list[string]
provenance:
authors:
- name: Huntbase hunt generation
org: huntbase.io
generated:
by: huntbase-hunt-generation
from: https://www.elastic.co/security-labs/blog/openshift-compliance-elastic-agent-builder
gates:
- dry-run
- lint
- critic
model: hb_google/gemini-3-flash-preview
rationale: Target clusters with high-severity failed scan results in hb_compliance_finding
first. Use the associated node hostnames to narrow the scope of the rare process
activity baseline.
references:
- name: 'From posture to root cause: Investigate OpenShift compliance with Elastic
Agent Builder'
url: https://www.elastic.co/security-labs/blog/openshift-compliance-elastic-agent-builder
related:
- hunt: kubernetes-rbac-tampering-investigation
reason: This hunt correlates scan results; general RBAC monitoring requires a broader
audit independent of compliance drift.
relation: out-of-scope-alternative
scenario:
stages:
- name: Unauthorized Cluster Resource Modification
observables:
- API calls to create or patch 'pods', 'serviceaccounts', or 'rolebindings'
- oc CLI commands targeting cluster configuration
- Deployment of workloads with specs violating CIS Benchmarks or NIST 800-53 moderate
controls
- metadata.name
- app.kubernetes.io
slug: unauthorized-resource-modification
tactic: execution
techniques:
- T1610
- T1059.004
- name: Compliance Violation Detection
observables:
- Resource results in 'compliancecheckresults.compliance.openshift.io'
- Status changes to 'FAIL' in 'compliancescans.compliance.openshift.io'
- Manual rule evaluation using Common Expression Language (CEL) custom rules
- oc get compliancecheckresult
- orchestrator.cluster.name
slug: compliance-violation-detection
tactic: discovery
techniques:
- T1613
- name: Root Cause Investigation and Remediation
observables:
- oc create token elastic-agent
- API activity involving 'complianceremediations.compliance.openshift.io'
- Correlation of 'actor_user_name' in Kubernetes audit logs with timestamps of
failed checks
- Automated annotation of manual rules via Elastic Workflows
- pods.items.all
- spec.id
slug: root-cause-investigation
tactic: impact
techniques:
- T1562
summary: An unauthorized or non-compliant configuration change in an OpenShift cluster
triggers a policy violation detected by the Compliance Operator. Defenders use
centralized telemetry to correlate these scan results with Kubernetes audit logs,
tracing the root cause back to the responsible principal and specific resource
modifications that caused the policy drift.
severity: medium
targets:
analyst:
name: Tier-2 analyst
role: analyst
endpoint:
category: endpoint
name: Endpoint telemetry (hb_ surfaces)
telemetry:
- endpoint
hunter:
agent: true
name: Hunt agent
tlp: clear
type: investigation
---
# OpenShift Compliance Drift and Audit Correlation
This hunt identifies security posture drift in OpenShift environments by correlating Compliance Operator scan results with Kubernetes API audit logs and node-level process telemetry. It begins with a scoping query of the compliance findings to identify failures, then gates a more resource-intensive analysis of audit logs and rare process executions to determine the root cause of the unauthorized modification.
## compliance-drift-lead
<!-- Lead: Identify Failed Compliance Controls -->
Find clusters and specific resources that have failed compliance checks according to the OpenShift Compliance Operator.
```sqlite target=endpoint role=scoping
~~~yaml
expected: Rows indicate failed CIS or NIST controls. Silence provides evidence that
the operator reports no active violations across the estate.
reads:
- resource_uid
- resource_name
- check_name
- status
- severity
- framework_refs
silence: evidence_of_absence
source: hb_compliance_finding
verified: dry-run
verified_at: '2026-11-20'
~~~
SELECT resource_uid, resource_name, check_name, status, severity, framework_refs FROM hb_compliance_finding WHERE LOWER(provider) = 'openshift' AND LOWER(status) = 'fail'
```
## analyze-compliance-status
<!-- Initial Assessment of Failures -->
```agent target=hunter
cite: required
context:
- compliance-drift-lead
max_iterations: 3
objective: Summarize which clusters and resources have high-severity failures and
provide a prioritized list for the gate decision.
success_criteria: A verdict on whether the failures justify audit correlation.
tools:
- endpoint
```
## gate-on-violations
<!-- Gate: Proceed to Audit? -->
if~: "the analysis shows active compliance failures on production clusters for sensitive resource types" (confidence: high, judge=hunter)
then: → audit-fan-out
indeterminate: → analyst-review
unavailable: → analyst-review (blind_spot: compliance-operator-down)
else: → close-out
## audit-fan-out
<!-- Corroborate with API and Process Activity -->
parallel:
- → audit-api-modifications
- → rare-cluster-ops
join: → triage-root-cause
## audit-api-modifications
<!-- Kubernetes API Resource Modifications -->
Find recent create or patch operations on sensitive resources that likely caused the compliance drift.
```sqlite target=endpoint role=detection-candidate params=(lookback_days=lookback_days, sensitive_resources=sensitive_resources)
~~~yaml
expected: API activity identifying the actor who modified the resource. Signal is
validated when it precedes the compliance scan failure.
reads:
- time
- actor_user_name
- api_operation
- resource_type
- resource_name
- cloud_account_uid
- src_endpoint_ip
silence: not_evidence_of_absence
source: hb_cloud_api_activity
verified: dry-run
verified_at: '2026-11-20'
~~~
SELECT time, actor_user_name, api_operation, resource_type, resource_name, cloud_account_uid, src_endpoint_ip FROM hb_cloud_api_activity WHERE provider = 'kubernetes' AND (LOWER(api_operation) = 'create' OR LOWER(api_operation) = 'patch') AND instr(',' || '{{sensitive_resources}}' || ',', ',' || LOWER(resource_type) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days')
```
## rare-cluster-ops
<!-- Rare CLI Tool Execution on Nodes -->
Identify rare interactive usage of cluster tools like oc or kubectl directly on nodes, indicating manual tampering.
```sqlite target=endpoint role=baseline params=(lookback_days=lookback_days, scope_hosts=scope_hosts)
~~~yaml
baseline:
compare: first_seen
window: '{{lookback_days}}d'
expected: A rare process execution path for a cluster CLI tool. Silence suggests normal
automated operations.
prevalence:
by: device_hostname
key:
- process_path
rare_below: 3
reads:
- device_hostname
- process_path
- process_name
- time
silence: not_evidence_of_absence
source: hb_process_activity
verified: dry-run
verified_at: '2026-11-20'
~~~
SELECT LOWER(process_path) AS path, COUNT(DISTINCT device_hostname) AS host_count, MIN(time) AS first_seen FROM hb_process_activity WHERE (LOWER(process_name) = 'oc' OR LOWER(process_name) = 'kubectl') AND ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || device_hostname || ',') > 0) AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY 1 HAVING host_count < 3
```
## triage-root-cause
<!-- Triage: Correlate Findings -->
```agent target=hunter
cite: required
context:
- analyze-compliance-status
- audit-api-modifications
- rare-cluster-ops
max_iterations: 6
objective: Determine if any resource modification in the audit results explains a
failed control found in the lead status assessment.
success_criteria: A per-cluster verdict identifying the actor and the infringing resource
change.
tools:
- endpoint
```
## route-remediation
<!-- Route to Remediation -->
if~: "the triage links a specific unauthorized modification to the timing of a compliance failure with high confidence" (confidence: high, judge=hunter)
then: → apply-remediation
indeterminate: → analyst-review
unavailable: → analyst-review (blind_spot: insufficient-audit-detail)
else: → close-out
## apply-remediation
<!-- Revert Resource Modification -->
```action target=endpoint
~~~yaml
approval: required
~~~
Revert the unauthorized pod, service account, or rolebinding configuration. If a compliance remediation object exists for the failed check, apply it.
```
→ analyst-review
## analyst-review
<!-- Analyst Review -->
```manual target=analyst
Verify the actor_user_name in audit logs against internal change requests. Ensure the scan status for the cluster moves back to PASS.
```
→ close-out
## close-out
<!-- Close Out -->
```manual target=analyst
Record the specific resource and actor that caused the drift. If the modification was a common vector, promote the audit query to a detection rule.
```
→ end
Run this hunt against your own data.
In Huntbase, Scout runs every step across the tools you already connect, checks back through your history and returns a verdict with the evidence. The hunt.md file is open, so you can also take it anywhere that reads the format.
Drafted by machine (huntbase-hunt-generation, hb_google/gemini-3-flash-preview), checked by dry-run, lint, critic, then reviewed by a person.